Weld Efficiency determination

[KiwiTanks]

We have taken a coupon out of a tank to determine the weld efficiency to use in the calculations and have the following test results:
508 MPa - UTS Parent metal (both Long and Trans)
410 MPa - UTS Weld 1
417 MPa - UTS Weld 2
375 MPa - UTS Weld 3

The cross sectional area of the test specimen are
20 x 9.5 Parent metal Longitudinal
20.1 x 9.5 Parent metal Transverse
19.1 x 11.5 Weld 1
19.1 x 10.3 Weld 2
19.1 x 10.6 Weld 3

A visual of the weld shows a lack of penetration, but there is excessive reinforcecment, (which would make up for the lack of penetration)

The inspection agency have provided a calculation with load (kN) divided by the the width (mm) of the specimen ie
Sample Width Load Load/width (kN/cm)
PMT 20 96.5 48.3
PML 20.1 97 48.3
W1 19.1 90 47.1
W2 19.1 82 42.9
W3 19.1 76 39.8

Then have used the average weld / average parent metal (see option c) below)


The question is How do you determine the weld efficiency of this tank? We have several answers, so I am trying to find which is betterer..

a) Lowest Weld UTS / Highest Parent UTS = 375/508 = 74%
b) Lowest Weld UTS / Average Parent UTS (in this case is the same) = 74%
c) Ave load/width of weld / Ave load/width of parent metal (as above) = 43.3/48.3 = 90%
d) Lowest load/width of weld / Ave load/width of parent metal = 39.8/48.3 = 82%
e) Something entirely different...???

Can anyone help? or point me to a document to read?

 

[JStephen]

Were all of these test specimens taken from one sample? That is, were they all from one fairly small area? And are they from a vertical weld?

My thinking would be to do some radiography and try to confirm if the lack of penetration is typical of all the verts, or if you just happened to hit a bad area. If it is typical, I'd aim for the lower end of your joint efficiency range. And I don't think there's going to be any reasonable way to get an "exact" number out of it.

In some of the older standards, round seams weren't required to be full penetration, so if you cut these out of a round seam, it wouldn't mean much.

As applied to tanks, the joint efficency is used to acknowledge the possibility of weld flaws, not to account for known lack of penetration. So to my way of thinking, if those samples are typical of the tank welding, I'd want to use a joint efficiency lower than calculated from the samples, realizing you might not have hit the worst spot with those samples.

Do you know what the material specification is? You could also base your joint efficiency on the minimum specified tensile strength, rather than the measured tensile strength.

 

[KiwiTanks]

Thanks for your comments.

We have taken these samples from a single plate about 500x700 cut out of a vertical weld. We are going to pick some of the worst looking welds, perform more radiography, cut out some more coupons from the worst places, and get them tested too.

I think that the info above will be a pretty good representative sample of what the tank welding will be like. (It was constructed in 1971 and looks like it's been kicked around a bit!) Check out the photo of the cross-section macro)

We have also performed an optical emmission spectroscopy on the parent steel to see if we can determine what the parent metal was. Still waiting to determine this - again it may be a bit of a guess.

At the moment that tank still stacks up using the API653 default materials, but further down the track, when the tank is another 40 years older, it may be a different story. Therefore the Client wants to use the 'best' results

Regards, KiwiTanks

 

[yosamitysamm]

The joint efficiency you end up using should contain 2 factors.
1. The joint efficiency of the weld as determined by your destructive testing and
2. The joint efficiency due to the reliability of the weld. API 620 would suggest that you use a joint efficiency of 0.85 for a double welded butt joint with spot RT (Table 5-2).

An average value for the joint efficiency in the weld would seem appropriate for 1 and then the efficiency for 2 covers off your lesser quality welds.


E of 0.82 x 0.85 = 0.70 This is still covering off your low end welds that have a E of 0.74.